Abstract. Detailed investigations of landslides are essential to understand fundamental landslide mechanisms. Seismic refraction method has been proven as a useful geophysical tool for investigating shallow landslides. The objective of this study is to introduce a new workflow using neural network in analyzing seismic refraction data and to compare the result with some methods; that are general reciprocal method (GRM) and refraction tomography. The GRM is effective when the velocity structure is relatively simple and refractors are gently dipping. Refraction tomography is capable of modeling the complex velocity structures of landslides. Neural network is found to be more potential in application especially in time consuming and complicated numerical methods. Neural network seem to have the ability to establish a relationship between an input and output space for mapping seismic velocity. Therefore, we made a preliminary attempt to evaluate the applicability of neural network to determine velocity and elevation of subsurface synthetic models corresponding to arrival times. The training and testing process of the neural network is successfully accomplished using the synthetic data. Furthermore, we evaluated the neural network using observed data. The result of the evaluation indicates that the neural network can compute velocity and elevation corresponding to arrival times. The similarity of those models shows the success of neural network as a new alternative in seismic refraction data interpretation.
Monitoring stability of the field in research area using time-lapse microgravity method has been done with data acquisition process in 2014 and 2016. This method requires high accuracy in µGal units. Since we use relative gravimeter, drift performance from instrument is also influential to the data significantly. In this paper we conducted reprocessing data using closer reference station from research area thus the drift effect cause by mobilitation time and duration of instrumental resting can be reduced. In conditions prior to injection, ideally the change in gravity value in the area would be very small with the assumption of the absence of injection activity. However, after the processing data stage we can see time-lapse microgravity anomalies within the range of ± 80 µGal. This anomaly was assumed occurred by instrument effect (typical for spring sensors) in field measurement and shallow hidrology effect (ground water level change). From the gravity surveys in 2014 and 2016, there is big difference in long term drift between 2014 and 2016 data, which is 2014 has average value of 320 µGal/day and 2016 has average value of 120 µGal/day. Other than that, in repeatability test, 2014 has 54% and 2016 has 77% from total data that is within the range of deviation ± 10 µGal (repeatability value in accuracy of ideal Scintrex CG-5). Analysis of gravity data measured in three benchmark (BM) was also performed to know the gravity variation rate in research area.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.